In an odd way, I'd like to see Liberty land the NY6 bid from the G5. This board would be abuzz.

(11-01-2023 02:17 PM)quo vadis Wrote:  

(11-01-2023 12:08 PM)JSchmack Wrote:  

(11-01-2023 09:15 AM)quo vadis Wrote:  Seems like the CFP might be just disregarding the Tulane loss to Ole Miss, and evaluating the two teams as if that game doesn't exist.

If so, Tulane might indeed be slightly better than AFA. Of course, no other ranking system does that, nor should the CFP IMO.

Part of the explanation for "ignoring" the Ole Miss loss is that:
(a) Air Force hasn't played a top 20 team; and
(b) Tulane's loss to Ole Miss was with their starting QB in street clothes; and the backup kept it close for 54 minutes (20-17) before turning it over on downs, fumbling it (returned for TD) and turning it over downs again.


The more interesting question is where would Tulane be ranked in the CFP if they were undefeated with a win over Ole Miss?

If the committee "forgives" a loss with a backup QB to put Tulane ahead of Air Force; but they're really just dumping Tulane and Air. Force at the bottom because they're G5 and not P5.

So they'll say "Tulane hasn't played anyone." but then rank a two-loss USC team with a WORSE SOS ahead of Tulane:

Michigan SOS is .5536 (31-25), #65 in FBS
Oregon SOS is .5536 (31-25), #65 in FBS
Louisville SOS is .5439 (31-26), #69 in FBS
Tulane SOS is .5273 (29-26), #76 in FBS
USC SOS is .5077 (33-32), #87 in FBS
Georgia SOS is .4912 (28-29), #98 in FBS


They just cherry pick to justify putting teams where they want them.

FWIW, I like SOS ratings done by computers like Sagarin, as they typically take a deeper dive. For example, if Team X plays 12-0 Georgia and Team Y plays 12-0 James Madison, IIRC in the NCAA SOS ratings Team X and Team Y would have equal SOS, because both played a 12-0 team. But we know Georgia plays a tougher schedule than James Madison, so playing Georgia should be a tougher SOS thing than playing James Madison.

In this case, Sagarin has Tulane's SOS at #96, while USC's is #46.

Very true. I like ESPNs resume page because it shows SOS and remaining SOS. It has AF at 131st in front of only FIU and Liberty. Their remaining SOS is 90th.

(11-01-2023 03:14 PM)WhoseHouse? Wrote:  

(11-01-2023 02:17 PM)quo vadis Wrote:  

(11-01-2023 12:08 PM)JSchmack Wrote:  

(11-01-2023 09:15 AM)quo vadis Wrote:  Seems like the CFP might be just disregarding the Tulane loss to Ole Miss, and evaluating the two teams as if that game doesn't exist.

If so, Tulane might indeed be slightly better than AFA. Of course, no other ranking system does that, nor should the CFP IMO.

Part of the explanation for "ignoring" the Ole Miss loss is that:
(a) Air Force hasn't played a top 20 team; and
(b) Tulane's loss to Ole Miss was with their starting QB in street clothes; and the backup kept it close for 54 minutes (20-17) before turning it over on downs, fumbling it (returned for TD) and turning it over downs again.


The more interesting question is where would Tulane be ranked in the CFP if they were undefeated with a win over Ole Miss?

If the committee "forgives" a loss with a backup QB to put Tulane ahead of Air Force; but they're really just dumping Tulane and Air. Force at the bottom because they're G5 and not P5.

So they'll say "Tulane hasn't played anyone." but then rank a two-loss USC team with a WORSE SOS ahead of Tulane:

Michigan SOS is .5536 (31-25), #65 in FBS
Oregon SOS is .5536 (31-25), #65 in FBS
Louisville SOS is .5439 (31-26), #69 in FBS
Tulane SOS is .5273 (29-26), #76 in FBS
USC SOS is .5077 (33-32), #87 in FBS
Georgia SOS is .4912 (28-29), #98 in FBS


They just cherry pick to justify putting teams where they want them.

FWIW, I like SOS ratings done by computers like Sagarin, as they typically take a deeper dive. For example, if Team X plays 12-0 Georgia and Team Y plays 12-0 James Madison, IIRC in the NCAA SOS ratings Team X and Team Y would have equal SOS, because both played a 12-0 team. But we know Georgia plays a tougher schedule than James Madison, so playing Georgia should be a tougher SOS thing than playing James Madison.

In this case, Sagarin has Tulane's SOS at #96, while USC's is #46.

Very true. I like ESPNs resume page because it shows SOS and remaining SOS. It has AF at 131st in front of only FIU and Liberty. Their remaining SOS is 90th.

yep, and thats the rub when looking at SOS numbers; what data actually makes up the sos in any given system.

Some will only use the win-loss record. THis is the way basketball rpi's sos metric was calculated back in the day (It may have changed since then) They basically take a team's opponent record, and their opponents opponents records to calculate a teams sos. Its a very shoddy way of doing it, and it really isnt gauging opponent's strength at all.

The better systems utilize a combo of win-loss records and offensive/defensive efficiencies, and other metrics which can drill into a teams' overall strength much more accurately.

(11-01-2023 04:16 PM)UofMstateU Wrote:  

(11-01-2023 03:14 PM)WhoseHouse? Wrote:  

(11-01-2023 02:17 PM)quo vadis Wrote:  FWIW, I like SOS ratings done by computers like Sagarin, as they typically take a deeper dive. For example, if Team X plays 12-0 Georgia and Team Y plays 12-0 James Madison, IIRC in the NCAA SOS ratings Team X and Team Y would have equal SOS, because both played a 12-0 team. But we know Georgia plays a tougher schedule than James Madison, so playing Georgia should be a tougher SOS thing than playing James Madison.

In this case, Sagarin has Tulane's SOS at #96, while USC's is #46.

Very true. I like ESPNs resume page because it shows SOS and remaining SOS. It has AF at 131st in front of only FIU and Liberty. Their remaining SOS is 90th.

yep, and thats the rub when looking at SOS numbers; what data actually makes up the sos in any given system.

Some will only use the win-loss record. THis is the way basketball rpi's sos metric was calculated back in the day (It may have changed since then) They basically take a team's opponent record, and their opponents opponents records to calculate a teams sos. Its a very shoddy way of doing it, and it really isnt gauging opponent's strength at all.

The better systems utilize a combo of win-loss records and offensive/defensive efficiencies, and other metrics which can drill into a teams' overall strength much more accurately.

The two problems with SOS based systems are:

#1 - We're talking about the top 20 percent of teams and using the decimal place differences in the records of the bottom 60% of college football... as if that matters at all. If you're in the Top 25 of the CFP Rankings... you're SUPPOSED TO BEAT all those teams in the bottom 60%. Why split hairs over whether you beat 2-7 Vandy or 1-8 Southern Miss?

#2 - There's simply no way to account for conference effect and the "two realities" of the P5 and the G5 (and this holds true in basketball ranking systems).

Tulane is 6-0 vs teams no one would consider Top 35 teams.
Kansas State is 6-0 vs teams no one considers Top 35 teams.
Oklahoma State is 4-2 vs teams no one considers Top 35 teams.

Then Oklahoma State beats Kansas State, and now they have a marquee win that shows you how good they really are! K-State is a top 30 team. That's better than any win Tulane has!

But K-State is a "marquee win" for Oklahoma State because they're 6-0 vs teams outside the Top 35 -- which is exactly what Tulane is!

K-State's "best win" is Troy (6-2, not receiving votes); while Memphis (6-2, receiving votes) is Tulane's best win.

Which means, the mindset is that "Tulane hasn't done enough to be considered good." but Oklahoma State is considered good for beating a team that hasn't don't enough to be considered good?

This is the logical fallacy of how we just non-P5 teams. If beating a schedule like Tulane's faced was so easy, then Oklahoma State would be 8-0, not 6-2 with losses to South Alabama (whom Tulane beat) and Iowa State (who's getting less votes than Memphis).

(11-02-2023 01:40 PM)JSchmack Wrote:  

(11-01-2023 04:16 PM)UofMstateU Wrote:  

(11-01-2023 03:14 PM)WhoseHouse? Wrote:  

(11-01-2023 02:17 PM)quo vadis Wrote:  FWIW, I like SOS ratings done by computers like Sagarin, as they typically take a deeper dive. For example, if Team X plays 12-0 Georgia and Team Y plays 12-0 James Madison, IIRC in the NCAA SOS ratings Team X and Team Y would have equal SOS, because both played a 12-0 team. But we know Georgia plays a tougher schedule than James Madison, so playing Georgia should be a tougher SOS thing than playing James Madison.

In this case, Sagarin has Tulane's SOS at #96, while USC's is #46.

Very true. I like ESPNs resume page because it shows SOS and remaining SOS. It has AF at 131st in front of only FIU and Liberty. Their remaining SOS is 90th.

yep, and thats the rub when looking at SOS numbers; what data actually makes up the sos in any given system.

Some will only use the win-loss record. THis is the way basketball rpi's sos metric was calculated back in the day (It may have changed since then) They basically take a team's opponent record, and their opponents opponents records to calculate a teams sos. Its a very shoddy way of doing it, and it really isnt gauging opponent's strength at all.

The better systems utilize a combo of win-loss records and offensive/defensive efficiencies, and other metrics which can drill into a teams' overall strength much more accurately.

The two problems with SOS based systems are:

#1 - We're talking about the top 20 percent of teams and using the decimal place differences in the records of the bottom 60% of college football... as if that matters at all. If you're in the Top 25 of the CFP Rankings... you're SUPPOSED TO BEAT all those teams in the bottom 60%. Why split hairs over whether you beat 2-7 Vandy or 1-8 Southern Miss?

#2 - There's simply no way to account for conference effect and the "two realities" of the P5 and the G5 (and this holds true in basketball ranking systems).

Tulane is 6-0 vs teams no one would consider Top 35 teams.
Kansas State is 6-0 vs teams no one considers Top 35 teams.
Oklahoma State is 4-2 vs teams no one considers Top 35 teams.

Then Oklahoma State beats Kansas State, and now they have a marquee win that shows you how good they really are! K-State is a top 30 team. That's better than any win Tulane has!

But K-State is a "marquee win" for Oklahoma State because they're 6-0 vs teams outside the Top 35 -- which is exactly what Tulane is!

K-State's "best win" is Troy (6-2, not receiving votes); while Memphis (6-2, receiving votes) is Tulane's best win.

Which means, the mindset is that "Tulane hasn't done enough to be considered good." but Oklahoma State is considered good for beating a team that hasn't don't enough to be considered good?

This is the logical fallacy of how we just non-P5 teams. If beating a schedule like Tulane's faced was so easy, then Oklahoma State would be 8-0, not 6-2 with losses to South Alabama (whom Tulane beat) and Iowa State (who's getting less votes than Memphis).

And none of it counts for the intangibles. There are teams that just know how to win and teams that just know how to lose. The strongest, most talented team is not always the best "team."

Power schools with 2 or fewer losses, unranked:
Iowa
Miami
North Carolina
Rutgers

Non-power schools with 1 or fewer losses, unranked:
Fresno St
* James Madison
Liberty
Toledo

All 25 ranked schools meet the above criteria.